Predicting cognitive load in immersive driving scenarios with a hybrid CNN-RNN model
archive: archived pipeline: cataloged verified
Get this paper ↗ (full text — opens at the source; we link to it, we don't host it)
Summary
This study addresses the prediction of cognitive load in drivers performing secondary tasks under challenging environmental conditions, specifically nighttime and rainy weather. While previous research often focused on moderate or low cognitive loads in monotonous, traffic-free settings, this work targets high cognitive load scenarios using an auditory n-back task (0-back, 1-back, and 2-back levels) as a secondary task. The research aims to improve the accuracy of cognitive load assessment by integrating functional near-infrared spectroscopy (fNIRS), eye-tracking, and driving behavior data, utilizing a novel hybrid neural network architecture. The experimental design involved 10 healthy adults driving in a simulator equipped with a motion platform and using Euro Truck Simulator 2 software to replicate low-visibility conditions. Data collection included fNIRS signals from 204 channels, eye-tracking metrics, and driving parameters such as speed, steering angle, throttle, and braking responses. The authors proposed a hybrid model combining a 1D Convolutional Neural Network (CNN) for spatial feature extraction and a Recurrent Neural Network (RNN) for capturing temporal dependencies. To optimize input features, the study compared several dimensionality reduction techniques, including Principal Component Analysis (PCA), Analysis of Variance (ANOVA), variance thresholding, and the Extra Trees classifier. The Extra Trees method was selected for its ability to identify the most relevant features while reducing model complexity. Results demonstrated that the proposed CNN-RNN model significantly outperformed previous methods and alternative feature selection techniques. When using combined physiological and driving data, the model achieved an accuracy of 99.91% with the Extra Trees feature selection, surpassing the 99.60% accuracy obtained with ANOVA and the 90.60% with PCA. Notably, the model improved accuracy from 99.82% to 99.91% using physiological data alone and from 87.26% to 92.02% using only driving behavior data, compared to a previous Conv-LSTM approach. Feature analysis revealed that fNIRS oxygenation channels and driving metrics like car speed and throttle response were the most critical predictors. Furthermore, the study found that features from different sensor modalities were largely independent, validating the benefit of multi-modal integration. The significance of this work lies in its demonstration that a hybrid CNN-RNN architecture, paired with Extra Trees feature selection, can accurately predict high cognitive load in realistic, adverse driving conditions with fewer parameters than previous models. This approach enhances the potential for developing robust Advanced Driver Assistance Systems (ADAS) capable of monitoring driver state in real-time during complex, high-risk scenarios. The findings underscore the importance of considering environmental factors and utilizing multi-modal data to capture the nuanced dynamics of driver cognitive workload.
Key finding
A lighter hybrid 1D-CNN + RNN improved three-level cognitive-load classification to 99.99% with physiological signals and 92.02% with driving-behavior signals, surpassing prior CNN-LSTM baselines.
Methodology
simulator
Sample size: N=10 (9 male, 1 female)
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed. Discovered via discover_arxiv on 2026-05-04 (4 acquisition events logged).
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | arxiv | — | — | 3 | 2026-05-04 |
| archive | success | — | — | — | 1 | 2026-05-04 |
| extract | success | cached | — | — | 2 | 2026-06-10 |
| clean | success | — | — | — | 1 | 2026-06-01 |
| chunk | success | — | — | — | 1 | 2026-06-01 |
| embed | success | — | — | — | 1 | 2026-06-02 |
| enrich | success | — | — | — | 1 | 2026-05-04 |
| promote | success | — | — | — | 1 | 2026-05-04 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 2 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 17 | 2026-06-11 |
| verify | success | — | — | — | 2 | 2026-06-10 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
- workload measurement
- mental demand
- distraction detection algorithms
- drowsiness detection algorithms
- road complexity
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Empirical Findings: physiological data
- Methodological Resource: tool software
- Theoretical Contribution: computational model